Combining Reinforcement Learning with Model Predictive Control for On-Ramp Merging

This repository contains the code used for the experiments in our paper of the above name, along with some experimental results (soon!)

Setup instructions:

1. Install SUMO version 1.7.0 and set SUMO_HOME to point to that installation
2. Install python 3.6 or above
3. Clone this repository
4. (If desired) Create a virtual environment for this project and activate it

   python3 -m venv ./mergeEnv/
   source mergeEnv/bin/activate

5. (If CUDA support is desired) Install PyTorch with CUDA support, following the instructions on the PyTorch website.
6. Install the other requirements with pip

   cd RL-MPC-LaneMerging
   pip install -r requirements.txt

7. Compile the Cython: python setup.py build_ext --inplace

Running experiments:

Run main.py, using the name of the desired configuration file as an argument. All configuration files used for our experiments can be found in the configs folder.

python main.py configs/<settings>.json

The results of each experiment will be appended to a file named run_data.csv.

Full experimental results from the paper:

See the experiment_data folder